Is it worthwhile to rebuild existing software for learning?

It’s definitely not dumb if your goal is to understand how things work. Rebuilding familiar tools gives you hands-on practice and can expose design decisions that aren’t obvious from the outside. Just compare your version with established implementations afterward so you can see what they handle that yours doesn’t.

A good rule is: rebuild things when you want to learn, and use existing tools when you need to finish a real job efficiently. Both approaches have a place. There’s also a middle ground—make a small educational version, then study established projects to learn why their designs are more complicated.

A simpler implementation can be useful, but make sure you understand what requirements you’re leaving out. An email service may need to queue messages, store and forward them when a recipient is offline, handle retries and failures, authenticate users, and deal with systems that aren’t online at the same time. Simplifying the implementation is fine when it’s an intentional tradeoff, not when important requirements are being overlooked.

The distinction is whether you’re learning or trying to ship a reliable product. For practice, recreating an email system is a great project. For production, existing libraries and services are usually safer, more capable, and much better tested. Be especially careful with security, encryption, authentication, and anything that could expose user data—those are areas where writing your own replacement can create serious problems.